kengine

package module
v1.0.4 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Sep 29, 2024 License: Apache-2.0 Imports: 56 Imported by: 0

README

Kengine

a project

Every site on HTTPS

Kengine is an extensible server platform that uses TLS by default.


Kengine Forum
Kengine on Sourcegraph Cloudsmith

Releases · Documentation · Get Help

Menu

Powered by
CertMagic

Features

  • Easy configuration with the Kenginefile
  • Powerful configuration with its native JSON config
  • Dynamic configuration with the JSON API
  • Config adapters if you don't like JSON
  • Automatic HTTPS by default
    • ZeroSSL and Let's Encrypt for public names
    • Fully-managed local CA for internal names & IPs
    • Can coordinate with other Kengine instances in a cluster
    • Multi-issuer fallback
  • Stays up when other servers go down due to TLS/OCSP/certificate-related issues
  • Production-ready after serving trillions of requests and managing millions of TLS certificates
  • Scales to hundreds of thousands of sites as proven in production
  • HTTP/1.1, HTTP/2, and HTTP/3 all supported by default
  • Highly extensible modular architecture lets Kengine do anything without bloat
  • Runs anywhere with no external dependencies (not even libc)
  • Written in Go, a language with higher memory safety guarantees than other servers
  • Actually fun to use
  • So much more to discover

Install

The simplest, cross-platform way to get started is to download Kengine from GitHub Releases and place the executable file in your PATH.

See our online documentation for other install instructions.

Build from source

Requirements:

For development

Note: These steps will not embed proper version information. For that, please follow the instructions in the next section.

$ git clone "https://github.com/khulnasoft/kengine.git"
$ cd kengine/cmd/kengine/
$ go build

When you run Kengine, it may try to bind to low ports unless otherwise specified in your config. If your OS requires elevated privileges for this, you will need to give your new binary permission to do so. On Linux, this can be done easily with: sudo setcap cap_net_bind_service=+ep ./kengine

If you prefer to use go run which only creates temporary binaries, you can still do this with the included setcap.sh like so:

$ go run -exec ./setcap.sh main.go

If you don't want to type your password for setcap, use sudo visudo to edit your sudoers file and allow your user account to run that command without a password, for example:

username ALL=(ALL:ALL) NOPASSWD: /usr/sbin/setcap

replacing username with your actual username. Please be careful and only do this if you know what you are doing! We are only qualified to document how to use Kengine, not Go tooling or your computer, and we are providing these instructions for convenience only; please learn how to use your own computer at your own risk and make any needful adjustments.

With version information and/or plugins

Using our builder tool, xkengine...

$ xkengine build

...the following steps are automated:

  1. Create a new folder: mkdir kengine
  2. Change into it: cd kengine
  3. Copy Kengine's main.go into the empty folder. Add imports for any custom plugins you want to add.
  4. Initialize a Go module: go mod init kengine
  5. (Optional) Pin Kengine version: go get github.com/khulnasoft/kengine@version replacing version with a git tag, commit, or branch name.
  6. (Optional) Add plugins by adding their import: _ "import/path/here"
  7. Compile: go build

Quick start

The Kengine website has documentation that includes tutorials, quick-start guides, reference, and more.

We recommend that all users -- regardless of experience level -- do our Getting Started guide to become familiar with using Kengine.

If you've only got a minute, the website has several quick-start tutorials to choose from! However, after finishing a quick-start tutorial, please read more documentation to understand how the software works. 🙂

Overview

Kengine is most often used as an HTTPS server, but it is suitable for any long-running Go program. First and foremost, it is a platform to run Go applications. Kengine "apps" are just Go programs that are implemented as Kengine modules. Two apps -- tls and http -- ship standard with Kengine.

Kengine apps instantly benefit from automated documentation, graceful on-line config changes via API, and unification with other Kengine apps.

Although JSON is Kengine's native config language, Kengine can accept input from config adapters which can essentially convert any config format of your choice into JSON: Kenginefile, JSON 5, YAML, TOML, NGINX config, and more.

The primary way to configure Kengine is through its API, but if you prefer config files, the command-line interface supports those too.

Kengine exposes an unprecedented level of control compared to any web server in existence. In Kengine, you are usually setting the actual values of the initialized types in memory that power everything from your HTTP handlers and TLS handshakes to your storage medium. Kengine is also ridiculously extensible, with a powerful plugin system that makes vast improvements over other web servers.

To wield the power of this design, you need to know how the config document is structured. Please see our documentation site for details about Kengine's config structure.

Nearly all of Kengine's configuration is contained in a single config document, rather than being scattered across CLI flags and env variables and a configuration file as with other web servers. This makes managing your server config more straightforward and reduces hidden variables/factors.

Full documentation

Our website has complete documentation:

https://khulnasoft.com/docs/

The docs are also open source. You can contribute to them here: https://github.com/khulnasoft/website

Getting help

  • We advise companies using Kengine to secure a support contract through Ardan Labs before help is needed.

  • A sponsorship goes a long way! We can offer private help to sponsors. If Kengine is benefitting your company, please consider a sponsorship. This not only helps fund full-time work to ensure the longevity of the project, it provides your company the resources, support, and discounts you need; along with being a great look for your company to your customers and potential customers!

  • Individuals can exchange help for free on our community forum at https://kengine.community. Remember that people give help out of their spare time and good will. The best way to get help is to give it first!

Please use our issue tracker only for bug reports and feature requests, i.e. actionable development items (support questions will usually be referred to the forums).

About

Matthew Holt began developing Kengine in 2014 while studying computer science at Brigham Young University. (The name "Kengine" was chosen because this software helps with the tedious, mundane tasks of serving the Web, and is also a single place for multiple things to be organized together.) It soon became the first web server to use HTTPS automatically and by default, and now has hundreds of contributors and has served trillions of HTTPS requests.

The name "Kengine" is trademarked. The name of the software is "Kengine", not "Kengine Server" or "KhulnaSoft". Please call it "Kengine" or, if you wish to clarify, "the Kengine web server". Kengine is a registered trademark of Stack Holdings GmbH.

Kengine is a project of ZeroSSL, a Stack Holdings company.

Debian package repository hosting is graciously provided by Cloudsmith. Cloudsmith is the only fully hosted, cloud-native, universal package management solution, that enables your organization to create, store and share packages in any format, to any place, with total confidence.

Documentation

Index

Examples

Constants

View Source 
const (
	ExitCodeSuccess = iota
	ExitCodeFailedStartup
	ExitCodeForceQuit
	ExitCodeFailedQuit
)

Exit codes. Generally, you should NOT automatically restart the process if the exit code is ExitCodeFailedStartup (1).

View Source 
const DefaultLoggerName = "default"
View Source 
const ImportPath = "github.com/khulnasoft/kengine"

ImportPath is the package import path for Kengine core. This identifier may be removed in the future.

Variables

View Source 
var (
	// DefaultAdminListen is the address for the local admin
	// listener, if none is specified at startup.
	DefaultAdminListen = "localhost:2019"

	// DefaultRemoteAdminListen is the address for the remote
	// (TLS-authenticated) admin listener, if enabled and not
	// specified otherwise.
	DefaultRemoteAdminListen = ":2021"
)
View Source 
var ConfigAutosavePath = filepath.Join(AppConfigDir(), "autosave.json")

ConfigAutosavePath is the default path to which the last config will be persisted.

View Source 
var CustomVersion string

CustomVersion is an optional string that overrides Kengine's reported version. It can be helpful when downstream packagers need to manually set Kengine's version. If no other version information is available, the short form version (see Version()) will be set to CustomVersion, and the full version will include CustomVersion at the beginning.

Set this variable during `go build` with `-ldflags`: 

-ldflags '-X github.com/khulnasoft/kengine.CustomVersion=v2.6.2'

for example.

View Source 
var DefaultStorage = &certmagic.FileStorage{Path: AppDataDir()}

DefaultStorage is Kengine's default storage module.

View Source 
var ErrNotConfigured = fmt.Errorf("module not configured")

ErrNotConfigured indicates a module is not configured.

Functions

func AppConfigDir added in v1.0.4 

func AppConfigDir() string

AppConfigDir returns the directory where to store user's config.

If XDG_CONFIG_HOME is set, it returns: $XDG_CONFIG_HOME/kengine. Otherwise, os.UserConfigDir() is used; if successful, it appends "Kengine" (Windows & Mac) or "kengine" (every other OS) to the path. If it returns an error, the fallback path "./kengine" is returned.

The config directory is not guaranteed to be different from AppDataDir().

Unlike os.UserConfigDir(), this function prefers the XDG_CONFIG_HOME env var on all platforms, not just Unix.

Ref: https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html

func AppDataDir added in v1.0.4 

func AppDataDir() string

AppDataDir returns a directory path that is suitable for storing application data on disk. It uses the environment for finding the best place to store data, and appends a "kengine" or "Kengine" (depending on OS and environment) subdirectory.

For a base directory path: If XDG_DATA_HOME is set, it returns: $XDG_DATA_HOME/kengine; otherwise, on Windows it returns: %AppData%/Kengine, on Mac: $HOME/Library/Application Support/Kengine, on Plan9: $home/lib/kengine, on Android: $HOME/kengine, and on everything else: $HOME/.local/share/kengine.

If a data directory cannot be determined, it returns "./kengine" (this is not ideal, and the environment should be fixed).

The data directory is not guaranteed to be different from AppConfigDir().

Ref: https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html

func Exiting added in v1.0.4 

func Exiting() bool

Exiting returns true if the process is exiting. EXPERIMENTAL API: subject to change or removal.

func GetModuleID added in v1.0.4 

func GetModuleID(instance any) string

GetModuleID returns a module's ID from an instance of its value. If the value is not a module, an empty string will be returned.

func GetModuleName added in v1.0.4 

func GetModuleName(instance any) string

GetModuleName returns a module's name (the last label of its ID) from an instance of its value. If the value is not a module, an empty string will be returned.

func HomeDir added in v1.0.4 

func HomeDir() string

HomeDir returns the best guess of the current user's home directory from environment variables. If unknown, "." (the current directory) is returned instead, except GOOS=android, which returns "/sdcard".

func InstanceID added in v1.0.4 

func InstanceID() (uuid.UUID, error)

InstanceID returns the UUID for this instance, and generates one if it does not already exist. The UUID is stored in the local data directory, regardless of storage configuration, since each instance is intended to have its own unique ID.

func IsUnixNetwork added in v1.0.4 

func IsUnixNetwork(netw string) bool

IsUnixNetwork returns true if the netw is a unix network.

func IsWriterStandardStream added in v1.0.4 

func IsWriterStandardStream(wo WriterOpener) bool

IsWriterStandardStream returns true if the input is a writer-opener to a standard stream (stdout, stderr).

func JoinNetworkAddress added in v1.0.4 

func JoinNetworkAddress(network, host, port string) string

JoinNetworkAddress combines network, host, and port into a single address string of the form accepted by ParseNetworkAddress(). For unix sockets, the network should be "unix" (or "unixgram" or "unixpacket") and the path to the socket should be given as the host parameter.

func ListenerUsage added in v1.0.4 

func ListenerUsage(network, addr string) int

ListenerUsage returns the current usage count of the given listener address.

func Load added in v1.0.4 

func Load(cfgJSON []byte, forceReload bool) error

Load loads the given config JSON and runs it only if it is different from the current config or forceReload is true.

func Log added in v1.0.4 

func Log() *zap.Logger

Log returns the current default logger.

func Modules added in v1.0.4 

func Modules() []string

Modules returns the names of all registered modules in ascending lexicographical order.

func OnExit added in v1.0.4 

func OnExit(f func(context.Context))

OnExit registers a callback to invoke during process exit. This registration is PROCESS-GLOBAL, meaning that each function should only be registered once forever, NOT once per config load (etc).

EXPERIMENTAL API: subject to change or removal.

func PIDFile added in v1.0.4 

func PIDFile(filename string) error

PIDFile writes a pidfile to the file at filename. It will get deleted before the process gracefully exits.

func ParseDuration added in v1.0.4 

func ParseDuration(s string) (time.Duration, error)

ParseDuration parses a duration string, adding support for the "d" unit meaning number of days, where a day is assumed to be 24h. The maximum input string length is 1024.

func ParseStructTag added in v1.0.4 

func ParseStructTag(tag string) (map[string]string, error)

ParseStructTag parses a kengine struct tag into its keys and values. It is very simple. The expected syntax is: `kengine:"key1=val1 key2=val2 ..."`

func RegisterModule added in v1.0.4 

func RegisterModule(instance Module)

RegisterModule registers a module by receiving a plain/empty value of the module. For registration to be properly recorded, this should be called in the init phase of runtime. Typically, the module package will do this as a side-effect of being imported. This function panics if the module's info is incomplete or invalid, or if the module is already registered.

func RegisterNetwork added in v1.0.4 

func RegisterNetwork(network string, getListener ListenerFunc)

RegisterNetwork registers a network type with Kengine so that if a listener is created for that network type, getListener will be invoked to get the listener. This should be called during init() and will panic if the network type is standard or reserved, or if it is already registered. EXPERIMENTAL and subject to change.

func RemoveMetaFields added in v1.0.4 

func RemoveMetaFields(rawJSON []byte) []byte

RemoveMetaFields removes meta fields like "@id" from a JSON message by using a simple regular expression. (An alternate way to do this would be to delete them from the raw, map[string]any representation as they are indexed, then iterate the index we made and add them back after encoding as JSON, but this is simpler.)

func Run added in v1.0.4 

func Run(cfg *Config) error

Run runs the given config, replacing any existing config.

func SplitNetworkAddress added in v1.0.4 

func SplitNetworkAddress(a string) (network, host, port string, err error)

SplitNetworkAddress splits a into its network, host, and port components. Note that port may be a port range (:X-Y), or omitted for unix sockets.

func Stop 

func Stop() error

Stop stops running the current configuration. It is the antithesis of Run(). This function will log any errors that occur during the stopping of individual apps and continue to stop the others. Stop should only be called if not replacing with a new config.

func StrictUnmarshalJSON added in v1.0.4 

func StrictUnmarshalJSON(data []byte, v any) error

StrictUnmarshalJSON is like json.Unmarshal but returns an error if any of the fields are unrecognized. Useful when decoding module configurations, where you want to be more sure they're correct.

func ToString added in v1.0.4 

func ToString(val any) string

ToString returns val as a string, as efficiently as possible. EXPERIMENTAL: may be changed or removed later.

func TrapSignals 

func TrapSignals()

TrapSignals create signal/interrupt handlers as best it can for the current OS. This is a rather invasive function to call in a Go program that captures signals already, so in that case it would be better to implement these handlers yourself.

func Validate added in v1.0.4 

func Validate(cfg *Config) error

Validate loads, provisions, and validates cfg, but does not start running it.

func Version added in v1.0.4 

func Version() (simple, full string)

Version returns the Kengine version in a simple/short form, and a full version string. The short form will not have spaces and is intended for User-Agent strings and similar, but may be omitting valuable information. Note that Kengine must be compiled in a special way to properly embed complete version information. First this function tries to get the version from the embedded build info provided by go.mod dependencies; then it tries to get info from embedded VCS information, which requires having built Kengine from a git repository. If no version is available, this function returns "(devel)" because Go uses that, but for the simple form we change it to "unknown". If still no version is available (e.g. no VCS repo), then it will use CustomVersion; CustomVersion is always prepended to the full version string.

See relevant Go issues: https://github.com/golang/go/issues/29228 and https://github.com/golang/go/issues/50603.

This function is experimental and subject to change or removal.

Types

type APIError added in v1.0.4 

type APIError struct {
	HTTPStatus int    `json:"-"`
	Err        error  `json:"-"`
	Message    string `json:"error"`
}

APIError is a structured error that every API handler should return for consistency in logging and client responses. If Message is unset, then Err.Error() will be serialized in its place.

func (APIError) Error added in v1.0.4 

func (e APIError) Error() string

type AdminAccess added in v1.0.4 

type AdminAccess struct {
	// Base64-encoded DER certificates containing public keys to accept.
	// (The contents of PEM certificate blocks are base64-encoded DER.)
	// Any of these public keys can appear in any part of a verified chain.
	PublicKeys []string `json:"public_keys,omitempty"`

	// Limits what the associated identities are allowed to do.
	// If unspecified, all permissions are granted.
	Permissions []AdminPermissions `json:"permissions,omitempty"`
	// contains filtered or unexported fields
}

AdminAccess specifies what permissions an identity or group of identities are granted.

type AdminConfig added in v1.0.4 

type AdminConfig struct {
	// If true, the admin endpoint will be completely disabled.
	// Note that this makes any runtime changes to the config
	// impossible, since the interface to do so is through the
	// admin endpoint.
	Disabled bool `json:"disabled,omitempty"`

	// The address to which the admin endpoint's listener should
	// bind itself. Can be any single network address that can be
	// parsed by Kengine. Accepts placeholders.
	// Default: the value of the `KENGINE_ADMIN` environment variable,
	// or `localhost:2019` otherwise.
	//
	// Remember: When changing this value through a config reload,
	// be sure to use the `--address` CLI flag to specify the current
	// admin address if the currently-running admin endpoint is not
	// the default address.
	Listen string `json:"listen,omitempty"`

	// If true, CORS headers will be emitted, and requests to the
	// API will be rejected if their `Host` and `Origin` headers
	// do not match the expected value(s). Use `origins` to
	// customize which origins/hosts are allowed. If `origins` is
	// not set, the listen address is the only value allowed by
	// default. Enforced only on local (plaintext) endpoint.
	EnforceOrigin bool `json:"enforce_origin,omitempty"`

	// The list of allowed origins/hosts for API requests. Only needed
	// if accessing the admin endpoint from a host different from the
	// socket's network interface or if `enforce_origin` is true. If not
	// set, the listener address will be the default value. If set but
	// empty, no origins will be allowed. Enforced only on local
	// (plaintext) endpoint.
	Origins []string `json:"origins,omitempty"`

	// Options pertaining to configuration management.
	Config *ConfigSettings `json:"config,omitempty"`

	// Options that establish this server's identity. Identity refers to
	// credentials which can be used to uniquely identify and authenticate
	// this server instance. This is required if remote administration is
	// enabled (but does not require remote administration to be enabled).
	// Default: no identity management.
	Identity *IdentityConfig `json:"identity,omitempty"`

	// Options pertaining to remote administration. By default, remote
	// administration is disabled. If enabled, identity management must
	// also be configured, as that is how the endpoint is secured.
	// See the neighboring "identity" object.
	//
	// EXPERIMENTAL: This feature is subject to change.
	Remote *RemoteAdmin `json:"remote,omitempty"`
	// contains filtered or unexported fields
}

AdminConfig configures Kengine's API endpoint, which is used to manage Kengine while it is running.

type AdminHandler added in v1.0.4 

type AdminHandler interface {
	ServeHTTP(http.ResponseWriter, *http.Request) error
}

AdminHandler is like http.Handler except ServeHTTP may return an error.

If any handler encounters an error, it should be returned for proper handling.

type AdminHandlerFunc added in v1.0.4 

type AdminHandlerFunc func(http.ResponseWriter, *http.Request) error

AdminHandlerFunc is a convenience type like http.HandlerFunc.

func (AdminHandlerFunc) ServeHTTP added in v1.0.4 

func (f AdminHandlerFunc) ServeHTTP(w http.ResponseWriter, r *http.Request) error

ServeHTTP implements the Handler interface.

type AdminPermissions added in v1.0.4 

type AdminPermissions struct {
	// The API paths allowed. Paths are simple prefix matches.
	// Any subpath of the specified paths will be allowed.
	Paths []string `json:"paths,omitempty"`

	// The HTTP methods allowed for the given paths.
	Methods []string `json:"methods,omitempty"`
}

AdminPermissions specifies what kinds of requests are allowed to be made to the admin endpoint.

type AdminRoute added in v1.0.4 

type AdminRoute struct {
	Pattern string
	Handler AdminHandler
}

AdminRoute represents a route for the admin endpoint.

type AdminRouter added in v1.0.4 

type AdminRouter interface {
	Routes() []AdminRoute
}

AdminRouter is a type which can return routes for the admin API.

type App added in v1.0.4 

type App interface {
	Start() error
	Stop() error
}

App is a thing that Kengine runs.

type BaseLog added in v1.0.4 

type BaseLog struct {
	// The module that writes out log entries for the sink.
	WriterRaw json.RawMessage `json:"writer,omitempty" kengine:"namespace=kengine.logging.writers inline_key=output"`

	// The encoder is how the log entries are formatted or encoded.
	EncoderRaw json.RawMessage `json:"encoder,omitempty" kengine:"namespace=kengine.logging.encoders inline_key=format"`

	// Tees entries through a zap.Core module which can extract
	// log entry metadata and fields for further processing.
	CoreRaw json.RawMessage `json:"core,omitempty" kengine:"namespace=kengine.logging.cores inline_key=module"`

	// Level is the minimum level to emit, and is inclusive.
	// Possible levels: DEBUG, INFO, WARN, ERROR, PANIC, and FATAL
	Level string `json:"level,omitempty"`

	// Sampling configures log entry sampling. If enabled,
	// only some log entries will be emitted. This is useful
	// for improving performance on extremely high-pressure
	// servers.
	Sampling *LogSampling `json:"sampling,omitempty"`

	// If true, the log entry will include the caller's
	// file name and line number. Default off.
	WithCaller bool `json:"with_caller,omitempty"`

	// If non-zero, and `with_caller` is true, this many
	// stack frames will be skipped when determining the
	// caller. Default 0.
	WithCallerSkip int `json:"with_caller_skip,omitempty"`

	// If not empty, the log entry will include a stack trace
	// for all logs at the given level or higher. See `level`
	// for possible values. Default off.
	WithStacktrace string `json:"with_stacktrace,omitempty"`
	// contains filtered or unexported fields
}

BaseLog contains the common logging parameters for logging.

type CleanerUpper added in v1.0.4 

type CleanerUpper interface {
	Cleanup() error
}

CleanerUpper is implemented by modules which may have side-effects such as opened files, spawned goroutines, or allocated some sort of non-stack state when they were provisioned. This method should deallocate/cleanup those resources to prevent memory leaks. Cleanup should be fast and efficient. Cleanup should work even if Provision returns an error, to allow cleaning up from partial provisionings.

type Config added in v1.0.4 

type Config struct {
	Admin   *AdminConfig `json:"admin,omitempty"`
	Logging *Logging     `json:"logging,omitempty"`

	// StorageRaw is a storage module that defines how/where Kengine
	// stores assets (such as TLS certificates). The default storage
	// module is `kengine.storage.file_system` (the local file system),
	// and the default path
	// [depends on the OS and environment](/docs/conventions#data-directory).
	StorageRaw json.RawMessage `json:"storage,omitempty" kengine:"namespace=kengine.storage inline_key=module"`

	// AppsRaw are the apps that Kengine will load and run. The
	// app module name is the key, and the app's config is the
	// associated value.
	AppsRaw ModuleMap `json:"apps,omitempty" kengine:"namespace="`
	// contains filtered or unexported fields
}

Config is the top (or beginning) of the Kengine configuration structure. Kengine config is expressed natively as a JSON document. If you prefer not to work with JSON directly, there are [many config adapters](/docs/config-adapters) available that can convert various inputs into Kengine JSON.

Many parts of this config are extensible through the use of Kengine modules. Fields which have a json.RawMessage type and which appear as dots (•••) in the online docs can be fulfilled by modules in a certain module namespace. The docs show which modules can be used in a given place.

Whenever a module is used, its name must be given either inline as part of the module, or as the key to the module's value. The docs will make it clear which to use.

Generally, all config settings are optional, as it is Kengine convention to have good, documented default values. If a parameter is required, the docs should say so.

Go programs which are directly building a Config struct value should take care to populate the JSON-encodable fields of the struct (i.e. the fields with `json` struct tags) if employing the module lifecycle (e.g. Provision method calls).

type ConfigLoader added in v1.0.4 

type ConfigLoader interface {
	LoadConfig(Context) ([]byte, error)
}

ConfigLoader is a type that can load a Kengine config. If the return value is non-nil, it must be valid Kengine JSON; if nil or with non-nil error, it is considered to be a no-op load and may be retried later.

type ConfigSettings added in v1.0.4 

type ConfigSettings struct {
	// Whether to keep a copy of the active config on disk. Default is true.
	// Note that "pulled" dynamic configs (using the neighboring "load" module)
	// are not persisted; only configs that are pushed to Kengine get persisted.
	Persist *bool `json:"persist,omitempty"`

	// Loads a new configuration. This is helpful if your configs are
	// managed elsewhere and you want Kengine to pull its config dynamically
	// when it starts. The pulled config completely replaces the current
	// one, just like any other config load. It is an error if a pulled
	// config is configured to pull another config without a load_delay,
	// as this creates a tight loop.
	//
	// EXPERIMENTAL: Subject to change.
	LoadRaw json.RawMessage `json:"load,omitempty" kengine:"namespace=kengine.config_loaders inline_key=module"`

	// The duration after which to load config. If set, config will be pulled
	// from the config loader after this duration. A delay is required if a
	// dynamically-loaded config is configured to load yet another config. To
	// load configs on a regular interval, ensure this value is set the same
	// on all loaded configs; it can also be variable if needed, and to stop
	// the loop, simply remove dynamic config loading from the next-loaded
	// config.
	//
	// EXPERIMENTAL: Subject to change.
	LoadDelay Duration `json:"load_delay,omitempty"`
}

ConfigSettings configures the management of configuration.

type ConfiguresFormatterDefault added in v1.0.4 

type ConfiguresFormatterDefault interface {
	ConfigureDefaultFormat(WriterOpener) error
}

ConfiguresFormatterDefault is an optional interface that encoder modules can implement to configure the default format of their encoder. This is useful for encoders which nest an encoder, that needs to know the writer in order to determine the correct default.

type Constructor added in v1.0.4 

type Constructor func() (Destructor, error)

Constructor is a function that returns a new value that can destruct itself when it is no longer needed.

type Context 

type Context struct {
	context.Context
	// contains filtered or unexported fields
}

Context is a type which defines the lifetime of modules that are loaded and provides access to the parent configuration that spawned the modules which are loaded. It should be used with care and wrapped with derivation functions from the standard context package only if you don't need the Kengine specific features. These contexts are canceled when the lifetime of the modules loaded from it is over.

Use NewContext() to get a valid value (but most modules will not actually need to do this).

func ActiveContext added in v1.0.4 

func ActiveContext() Context

ActiveContext returns the currently-active context. This function is experimental and might be changed or removed in the future.

func NewContext added in v1.0.4 

func NewContext(ctx Context) (Context, context.CancelFunc)

NewContext provides a new context derived from the given context ctx. Normally, you will not need to call this function unless you are loading modules which have a different lifespan than the ones for the context the module was provisioned with. Be sure to call the cancel func when the context is to be cleaned up so that modules which are loaded will be properly unloaded. See standard library context package's documentation.

func ProvisionContext added in v1.0.4 

func ProvisionContext(newCfg *Config) (Context, error)

ProvisionContext creates a new context from the configuration and provisions storage and app modules. The function is intended for testing and advanced use cases only, typically `Run` should be use to ensure a fully functional kengine instance. EXPERIMENTAL: While this is public the interface and implementation details of this function may change.

func (Context) App added in v1.0.4 

func (ctx Context) App(name string) (any, error)

App returns the configured app named name. If that app has not yet been loaded and provisioned, it will be immediately loaded and provisioned. If no app with that name is configured, a new empty one will be instantiated instead. (The app module must still be registered.) This must not be called during the Provision/Validate phase to reference a module's own host app (since the parent app module is still in the process of being provisioned, it is not yet ready).

We return any type instead of the App type because it is NOT intended for the caller of this method to be the one to start or stop App modules. The caller is expected to assert to the concrete type.

func (Context) AppIfConfigured added in v1.0.4 

func (ctx Context) AppIfConfigured(name string) (any, error)

AppIfConfigured is like App, but it returns an error if the app has not been configured. This is useful when the app is required and its absence is a configuration error; or when the app is optional and you don't want to instantiate a new one that hasn't been explicitly configured. If the app is not in the configuration, the error wraps ErrNotConfigured.

func (*Context) Filesystems added in v1.0.4 

func (ctx *Context) Filesystems() FileSystems

Filesystems returns a ref to the FilesystemMap. EXPERIMENTAL: This API is subject to change.

func (Context) IdentityCredentials added in v1.0.4 

func (ctx Context) IdentityCredentials(logger *zap.Logger) ([]tls.Certificate, error)

IdentityCredentials returns this instance's configured, managed identity credentials that can be used in TLS client authentication.

func (Context) LoadModule added in v1.0.4 

func (ctx Context) LoadModule(structPointer any, fieldName string) (any, error)

LoadModule loads the Kengine module(s) from the specified field of the parent struct pointer and returns the loaded module(s). The struct pointer and its field name as a string are necessary so that reflection can be used to read the struct tag on the field to get the module namespace and inline module name key (if specified).

The field can be any one of the supported raw module types: json.RawMessage, []json.RawMessage, map[string]json.RawMessage, or []map[string]json.RawMessage. ModuleMap may be used in place of map[string]json.RawMessage. The return value's underlying type mirrors the input field's type: 

json.RawMessage              => any
[]json.RawMessage            => []any
[][]json.RawMessage          => [][]any
map[string]json.RawMessage   => map[string]any
[]map[string]json.RawMessage => []map[string]any

The field must have a "kengine" struct tag in this format: 

kengine:"key1=val1 key2=val2"

To load modules, a "namespace" key is required. For example, to load modules in the "http.handlers" namespace, you'd put: `namespace=http.handlers` in the Kengine struct tag.

The module name must also be available. If the field type is a map or slice of maps, then key is assumed to be the module name if an "inline_key" is NOT specified in the kengine struct tag. In this case, the module name does NOT need to be specified in-line with the module itself.

If not a map, or if inline_key is non-empty, then the module name must be embedded into the values, which must be objects; then there must be a key in those objects where its associated value is the module name. This is called the "inline key", meaning the key containing the module's name that is defined inline with the module itself. You must specify the inline key in a struct tag, along with the namespace: 

kengine:"namespace=http.handlers inline_key=handler"

This will look for a key/value pair like `"handler": "..."` in the json.RawMessage in order to know the module name.

To make use of the loaded module(s) (the return value), you will probably want to type-assert each 'any' value(s) to the types that are useful to you and store them on the same struct. Storing them on the same struct makes for easy garbage collection when your host module is no longer needed.

Loaded modules have already been provisioned and validated. Upon returning successfully, this method clears the json.RawMessage(s) in the field since the raw JSON is no longer needed, and this allows the GC to free up memory.

Example 
// this whole first part is just setting up for the example;
// note the struct tags - very important; we specify inline_key
// because that is the only way to know the module name
var ctx Context
myStruct := &struct {
	// This godoc comment will appear in module documentation.
	GuestModuleRaw json.RawMessage `json:"guest_module,omitempty" kengine:"namespace=example inline_key=name"`

	// this is where the decoded module will be stored; in this
	// example, we pretend we need an io.Writer but it can be
	// any interface type that is useful to you
	guestModule io.Writer
}{
	GuestModuleRaw: json.RawMessage(`{"name":"module_name","foo":"bar"}`),
}

// if a guest module is provided, we can load it easily
if myStruct.GuestModuleRaw != nil {
	mod, err := ctx.LoadModule(myStruct, "GuestModuleRaw")
	if err != nil {
		// you'd want to actually handle the error here
		// return fmt.Errorf("loading guest module: %v", err)
	}
	// mod contains the loaded and provisioned module,
	// it is now ready for us to use
	myStruct.guestModule = mod.(io.Writer)
}

// use myStruct.guestModule from now on

Output:
Example (Array) 
// this whole first part is just setting up for the example;
// note the struct tags - very important; we specify inline_key
// because that is the only way to know the module name
var ctx Context
myStruct := &struct {
	// This godoc comment will appear in module documentation.
	GuestModulesRaw []json.RawMessage `json:"guest_modules,omitempty" kengine:"namespace=example inline_key=name"`

	// this is where the decoded module will be stored; in this
	// example, we pretend we need an io.Writer but it can be
	// any interface type that is useful to you
	guestModules []io.Writer
}{
	GuestModulesRaw: []json.RawMessage{
		json.RawMessage(`{"name":"module1_name","foo":"bar1"}`),
		json.RawMessage(`{"name":"module2_name","foo":"bar2"}`),
	},
}

// since our input is []json.RawMessage, the output will be []any
mods, err := ctx.LoadModule(myStruct, "GuestModulesRaw")
if err != nil {
	// you'd want to actually handle the error here
	// return fmt.Errorf("loading guest modules: %v", err)
}
for _, mod := range mods.([]any) {
	myStruct.guestModules = append(myStruct.guestModules, mod.(io.Writer))
}

// use myStruct.guestModules from now on

Output:
Example (Map) 
// this whole first part is just setting up for the example;
// note the struct tags - very important; we don't specify
// inline_key because the map key is the module name
var ctx Context
myStruct := &struct {
	// This godoc comment will appear in module documentation.
	GuestModulesRaw ModuleMap `json:"guest_modules,omitempty" kengine:"namespace=example"`

	// this is where the decoded module will be stored; in this
	// example, we pretend we need an io.Writer but it can be
	// any interface type that is useful to you
	guestModules map[string]io.Writer
}{
	GuestModulesRaw: ModuleMap{
		"module1_name": json.RawMessage(`{"foo":"bar1"}`),
		"module2_name": json.RawMessage(`{"foo":"bar2"}`),
	},
}

// since our input is map[string]json.RawMessage, the output will be map[string]any
mods, err := ctx.LoadModule(myStruct, "GuestModulesRaw")
if err != nil {
	// you'd want to actually handle the error here
	// return fmt.Errorf("loading guest modules: %v", err)
}
for modName, mod := range mods.(map[string]any) {
	myStruct.guestModules[modName] = mod.(io.Writer)
}

// use myStruct.guestModules from now on

Output:

func (Context) LoadModuleByID added in v1.0.4 

func (ctx Context) LoadModuleByID(id string, rawMsg json.RawMessage) (any, error)

LoadModuleByID decodes rawMsg into a new instance of mod and returns the value. If mod.New is nil, an error is returned. If the module implements Validator or Provisioner interfaces, those methods are invoked to ensure the module is fully configured and valid before being used.

This is a lower-level method and will usually not be called directly by most modules. However, this method is useful when dynamically loading/unloading modules in their own context, like from embedded scripts, etc.

func (Context) Logger added in v1.0.4 

func (ctx Context) Logger(module ...Module) *zap.Logger

Logger returns a logger that is intended for use by the most recent module associated with the context. Callers should not pass in any arguments unless they want to associate with a different module; it panics if more than 1 value is passed in.

Originally, this method's signature was `Logger(mod Module)`, requiring that an instance of a Kengine module be passed in. However, that is no longer necessary, as the closest module most recently associated with the context will be automatically assumed. To prevent a sudden breaking change, this method's signature has been changed to be variadic, but we may remove the parameter altogether in the future. Callers should not pass in any argument. If there is valid need to specify a different module, please open an issue to discuss.

PARTIALLY DEPRECATED: The Logger(module) form is deprecated and may be removed in the future. Do not pass in any arguments.

func (Context) Module added in v1.0.4 

func (ctx Context) Module() Module

Module returns the current module, or the most recent one provisioned by the context.

func (Context) Modules added in v1.0.4 

func (ctx Context) Modules() []Module

Modules returns the lineage of modules that this context provisioned, with the most recent/current module being last in the list.

func (*Context) OnCancel added in v1.0.4 

func (ctx *Context) OnCancel(f func())

OnCancel executes f when ctx is canceled.

func (*Context) OnExit added in v1.0.4 

func (ctx *Context) OnExit(f func(context.Context))

OnExit executes f when the process exits gracefully. The function is only executed if the process is gracefully shut down while this context is active.

EXPERIMENTAL API: subject to change or removal.

func (Context) Slogger added in v1.0.4 

func (ctx Context) Slogger() *slog.Logger

Slogger returns a slog logger that is intended for use by the most recent module associated with the context.

func (Context) Storage added in v1.0.4 

func (ctx Context) Storage() certmagic.Storage

Storage returns the configured Kengine storage implementation.

func (*Context) WithValue added in v1.0.4 

func (ctx *Context) WithValue(key, value any) Context

WithValue returns a new context with the given key-value pair.

type CtxKey 

type CtxKey string

CtxKey is a value type for use with context.WithValue. 

const ReplacerCtxKey CtxKey = "replacer"

ReplacerCtxKey is the context key for a replacer.

type CustomLog added in v1.0.4 

type CustomLog struct {
	BaseLog

	// Include defines the names of loggers to emit in this
	// log. For example, to include only logs emitted by the
	// admin API, you would include "admin.api".
	Include []string `json:"include,omitempty"`

	// Exclude defines the names of loggers that should be
	// skipped by this log. For example, to exclude only
	// HTTP access logs, you would exclude "http.log.access".
	Exclude []string `json:"exclude,omitempty"`
}

CustomLog represents a custom logger configuration.

By default, a log will emit all log entries. Some entries will be skipped if sampling is enabled. Further, the Include and Exclude parameters define which loggers (by name) are allowed or rejected from emitting in this log. If both Include and Exclude are populated, their values must be mutually exclusive, and longer namespaces have priority. If neither are populated, all logs are emitted.

type Destructor added in v1.0.4 

type Destructor interface {
	Destruct() error
}

Destructor is a value that can clean itself up when it is deallocated.

type DiscardWriter added in v1.0.4 

type DiscardWriter struct{}

DiscardWriter discards all writes.

func (DiscardWriter) KengineModule added in v1.0.4 

func (DiscardWriter) KengineModule() ModuleInfo

KengineModule returns the Kengine module information.

func (DiscardWriter) OpenWriter added in v1.0.4 

func (DiscardWriter) OpenWriter() (io.WriteCloser, error)

OpenWriter returns io.Discard that can't be closed.

func (DiscardWriter) String added in v1.0.4 

func (DiscardWriter) String() string

func (DiscardWriter) WriterKey added in v1.0.4 

func (DiscardWriter) WriterKey() string

WriterKey returns a unique key representing discard.

type Duration added in v1.0.4 

type Duration time.Duration

Duration can be an integer or a string. An integer is interpreted as nanoseconds. If a string, it is a Go time.Duration value such as `300ms`, `1.5h`, or `2h45m`; valid units are `ns`, `us`/`µs`, `ms`, `s`, `m`, `h`, and `d`.

func (*Duration) UnmarshalJSON added in v1.0.4 

func (d *Duration) UnmarshalJSON(b []byte) error

UnmarshalJSON satisfies json.Unmarshaler.

type FileSystems added in v1.0.4 

type FileSystems interface {
	Register(k string, v fs.FS)
	Unregister(k string)
	Get(k string) (v fs.FS, ok bool)
	Default() fs.FS
}

type IdentityConfig added in v1.0.4 

type IdentityConfig struct {
	// List of names or IP addresses which refer to this server.
	// Certificates will be obtained for these identifiers so
	// secure TLS connections can be made using them.
	Identifiers []string `json:"identifiers,omitempty"`

	// Issuers that can provide this admin endpoint its identity
	// certificate(s). Default: ACME issuers configured for
	// ZeroSSL and Let's Encrypt. Be sure to change this if you
	// require credentials for private identifiers.
	IssuersRaw []json.RawMessage `json:"issuers,omitempty" kengine:"namespace=tls.issuance inline_key=module"`
	// contains filtered or unexported fields
}

IdentityConfig configures management of this server's identity. An identity consists of credentials that uniquely verify this instance; for example, TLS certificates (public + private key pairs).

type ListenerFunc added in v1.0.4 

type ListenerFunc func(ctx context.Context, network, addr string, cfg net.ListenConfig) (any, error)

ListenerFunc is a function that can return a listener given a network and address. The listeners must be capable of overlapping: with Kengine, new configs are loaded before old ones are unloaded, so listeners may overlap briefly if the configs both need the same listener. EXPERIMENTAL and subject to change.

type ListenerWrapper added in v1.0.4 

type ListenerWrapper interface {
	WrapListener(net.Listener) net.Listener
}

ListenerWrapper is a type that wraps a listener so it can modify the input listener's methods. Modules that implement this interface are found in the kengine.listeners namespace. Usually, to wrap a listener, you will define your own struct type that embeds the input listener, then implement your own methods that you want to wrap, calling the underlying listener's methods where appropriate.

type LogSampling added in v1.0.4 

type LogSampling struct {
	// The window over which to conduct sampling.
	Interval time.Duration `json:"interval,omitempty"`

	// Log this many entries within a given level and
	// message for each interval.
	First int `json:"first,omitempty"`

	// If more entries with the same level and message
	// are seen during the same interval, keep one in
	// this many entries until the end of the interval.
	Thereafter int `json:"thereafter,omitempty"`
}

LogSampling configures log entry sampling.

type Logging added in v1.0.4 

type Logging struct {
	// Sink is the destination for all unstructured logs emitted
	// from Go's standard library logger. These logs are common
	// in dependencies that are not designed specifically for use
	// in Kengine. Because it is global and unstructured, the sink
	// lacks most advanced features and customizations.
	Sink *SinkLog `json:"sink,omitempty"`

	// Logs are your logs, keyed by an arbitrary name of your
	// choosing. The default log can be customized by defining
	// a log called "default". You can further define other logs
	// and filter what kinds of entries they accept.
	Logs map[string]*CustomLog `json:"logs,omitempty"`
	// contains filtered or unexported fields
}

Logging facilitates logging within Kengine. The default log is called "default" and you can customize it. You can also define additional logs.

By default, all logs at INFO level and higher are written to standard error ("stderr" writer) in a human-readable format ("console" encoder if stdout is an interactive terminal, "json" encoder otherwise).

All defined logs accept all log entries by default, but you can filter by level and module/logger names. A logger's name is the same as the module's name, but a module may append to logger names for more specificity. For example, you can filter logs emitted only by HTTP handlers using the name "http.handlers", because all HTTP handler module names have that prefix.

Kengine logs (except the sink) are zero-allocation, so they are very high-performing in terms of memory and CPU time. Enabling sampling can further increase throughput on extremely high-load servers.

func (*Logging) Logger added in v1.0.4 

func (logging *Logging) Logger(mod Module) *zap.Logger

Logger returns a logger that is ready for the module to use.

type Module added in v1.0.4 

type Module interface {
	// This method indicates that the type is a Kengine
	// module. The returned ModuleInfo must have both
	// a name and a constructor function. This method
	// must not have any side-effects.
	KengineModule() ModuleInfo
}

Module is a type that is used as a Kengine module. In addition to this interface, most modules will implement some interface expected by their host module in order to be useful. To learn which interface(s) to implement, see the documentation for the host module. At a bare minimum, this interface, when implemented, only provides the module's ID and constructor function.

Modules will often implement additional interfaces including Provisioner, Validator, and CleanerUpper. If a module implements these interfaces, their methods are called during the module's lifespan.

When a module is loaded by a host module, the following happens: 1) ModuleInfo.New() is called to get a new instance of the module. 2) The module's configuration is unmarshaled into that instance. 3) If the module is a Provisioner, the Provision() method is called. 4) If the module is a Validator, the Validate() method is called. 5) The module will probably be type-asserted from 'any' to some other, more useful interface expected by the host module. For example, HTTP handler modules are type-asserted as kenginehttp.MiddlewareHandler values. 6) When a module's containing Context is canceled, if it is a CleanerUpper, its Cleanup() method is called.

type ModuleID added in v1.0.4 

type ModuleID string

ModuleID is a string that uniquely identifies a Kengine module. A module ID is lightly structured. It consists of dot-separated labels which form a simple hierarchy from left to right. The last label is the module name, and the labels before that constitute the namespace (or scope).

Thus, a module ID has the form: <namespace>.<name>

An ID with no dot has the empty namespace, which is appropriate for app modules (these are "top-level" modules that Kengine core loads and runs).

Module IDs should be lowercase and use underscores (_) instead of spaces.

Examples of valid IDs: - http - http.handlers.file_server - kengine.logging.encoders.json

func (ModuleID) Name added in v1.0.4 

func (id ModuleID) Name() string

Name returns the Name (last element) of a module ID.

func (ModuleID) Namespace added in v1.0.4 

func (id ModuleID) Namespace() string

Namespace returns the namespace (or scope) portion of a module ID, which is all but the last label of the ID. If the ID has only one label, then the namespace is empty.

type ModuleInfo added in v1.0.4 

type ModuleInfo struct {
	// ID is the "full name" of the module. It
	// must be unique and properly namespaced.
	ID ModuleID

	// New returns a pointer to a new, empty
	// instance of the module's type. This
	// method must not have any side-effects,
	// and no other initialization should
	// occur within it. Any initialization
	// of the returned value should be done
	// in a Provision() method (see the
	// Provisioner interface).
	New func() Module
}

ModuleInfo represents a registered Kengine module.

func GetModule added in v1.0.4 

func GetModule(name string) (ModuleInfo, error)

GetModule returns module information from its ID (full name).

func GetModules added in v1.0.4 

func GetModules(scope string) []ModuleInfo

GetModules returns all modules in the given scope/namespace. For example, a scope of "foo" returns modules named "foo.bar", "foo.loo", but not "bar", "foo.bar.loo", etc. An empty scope returns top-level modules, for example "foo" or "bar". Partial scopes are not matched (i.e. scope "foo.ba" does not match name "foo.bar").

Because modules are registered to a map under the hood, the returned slice will be sorted to keep it deterministic.

func (ModuleInfo) String added in v1.0.4 

func (mi ModuleInfo) String() string

type ModuleMap added in v1.0.4 

type ModuleMap map[string]json.RawMessage

ModuleMap is a map that can contain multiple modules, where the map key is the module's name. (The namespace is usually read from an associated field's struct tag.) Because the module's name is given as the key in a module map, the name does not have to be given in the json.RawMessage.

type NetworkAddress added in v1.0.4 

type NetworkAddress struct {
	// Should be a network value accepted by Go's net package or
	// by a plugin providing a listener for that network type.
	Network string

	// The "main" part of the network address is the host, which
	// often takes the form of a hostname, DNS name, IP address,
	// or socket path.
	Host string

	// For addresses that contain a port, ranges are given by
	// [StartPort, EndPort]; i.e. for a single port, StartPort
	// and EndPort are the same. For no port, they are 0.
	StartPort uint
	EndPort   uint
}

NetworkAddress represents one or more network addresses. It contains the individual components for a parsed network address of the form accepted by ParseNetworkAddress().

func ParseNetworkAddress added in v1.0.4 

func ParseNetworkAddress(addr string) (NetworkAddress, error)

ParseNetworkAddress parses addr into its individual components. The input string is expected to be of the form "network/host:port-range" where any part is optional. The default network, if unspecified, is tcp. Port ranges are inclusive.

Network addresses are distinct from URLs and do not use URL syntax.

func ParseNetworkAddressWithDefaults added in v1.0.4 

func ParseNetworkAddressWithDefaults(addr, defaultNetwork string, defaultPort uint) (NetworkAddress, error)

ParseNetworkAddressWithDefaults is like ParseNetworkAddress but allows the default network and port to be specified.

func (NetworkAddress) At added in v1.0.4 

func (na NetworkAddress) At(portOffset uint) NetworkAddress

At returns a NetworkAddress with a port range of just 1 at the given port offset; i.e. a NetworkAddress that represents precisely 1 address only.

func (NetworkAddress) Expand added in v1.0.4 

func (na NetworkAddress) Expand() []NetworkAddress

Expand returns one NetworkAddress for each port in the port range.

func (NetworkAddress) IsUnixNetwork added in v1.0.4 

func (na NetworkAddress) IsUnixNetwork() bool

IsUnixNetwork returns true if na.Network is unix, unixgram, or unixpacket.

func (NetworkAddress) JoinHostPort added in v1.0.4 

func (na NetworkAddress) JoinHostPort(offset uint) string

JoinHostPort is like net.JoinHostPort, but where the port is StartPort + offset.

func (NetworkAddress) Listen added in v1.0.4 

func (na NetworkAddress) Listen(ctx context.Context, portOffset uint, config net.ListenConfig) (any, error)

Listen is similar to net.Listen, with a few differences:

Listen announces on the network address using the port calculated by adding portOffset to the start port. (For network types that do not use ports, the portOffset is ignored.)

The provided ListenConfig is used to create the listener. Its Control function, if set, may be wrapped by an internally-used Control function. The provided context may be used to cancel long operations early. The context is not used to close the listener after it has been created.

Kengine's listeners can overlap each other: multiple listeners may be created on the same socket at the same time. This is useful because during config changes, the new config is started while the old config is still running. How this is accomplished varies by platform and network type. For example, on Unix, SO_REUSEPORT is set except on Unix sockets, for which the file descriptor is duplicated and reused; on Windows, the close logic is virtualized using timeouts. Like normal listeners, be sure to Close() them when you are done.

This method returns any type, as the implementations of listeners for various network types are not interchangeable. The type of listener returned is switched on the network type. Stream-based networks ("tcp", "unix", "unixpacket", etc.) return a net.Listener; datagram-based networks ("udp", "unixgram", etc.) return a net.PacketConn; and so forth. The actual concrete types are not guaranteed to be standard, exported types (wrapping is necessary to provide graceful reloads).

Unix sockets will be unlinked before being created, to ensure we can bind to it even if the previous program using it exited uncleanly; it will also be unlinked upon a graceful exit (or when a new config does not use that socket).

func (NetworkAddress) ListenAll added in v1.0.4 

func (na NetworkAddress) ListenAll(ctx context.Context, config net.ListenConfig) ([]any, error)

ListenAll calls Listen() for all addresses represented by this struct, i.e. all ports in the range. (If the address doesn't use ports or has 1 port only, then only 1 listener will be created.) It returns an error if any listener failed to bind, and closes any listeners opened up to that point.

func (NetworkAddress) ListenQUIC added in v1.0.4 

func (na NetworkAddress) ListenQUIC(ctx context.Context, portOffset uint, config net.ListenConfig, tlsConf *tls.Config) (http3.QUICEarlyListener, error)

ListenQUIC returns a quic.EarlyListener suitable for use in a Kengine module. The network will be transformed into a QUIC-compatible type (if unix, then unixgram will be used; otherwise, udp will be used).

NOTE: This API is EXPERIMENTAL and may be changed or removed.

func (NetworkAddress) PortRangeSize added in v1.0.4 

func (na NetworkAddress) PortRangeSize() uint

PortRangeSize returns how many ports are in pa's port range. Port ranges are inclusive, so the size is the difference of start and end ports plus one.

func (NetworkAddress) String added in v1.0.4 

func (na NetworkAddress) String() string

String reconstructs the address string for human display. The output can be parsed by ParseNetworkAddress(). If the address is a unix socket, any non-zero port will be dropped.

type Provisioner added in v1.0.4 

type Provisioner interface {
	Provision(Context) error
}

Provisioner is implemented by modules which may need to perform some additional "setup" steps immediately after being loaded. Provisioning should be fast (imperceptible running time). If any side-effects result in the execution of this function (e.g. creating global state, any other allocations which require garbage collection, opening files, starting goroutines etc.), be sure to clean up properly by implementing the CleanerUpper interface to avoid leaking resources.

type RemoteAdmin added in v1.0.4 

type RemoteAdmin struct {
	// The address on which to start the secure listener. Accepts placeholders.
	// Default: :2021
	Listen string `json:"listen,omitempty"`

	// List of access controls for this secure admin endpoint.
	// This configures TLS mutual authentication (i.e. authorized
	// client certificates), but also application-layer permissions
	// like which paths and methods each identity is authorized for.
	AccessControl []*AdminAccess `json:"access_control,omitempty"`
}

RemoteAdmin enables and configures remote administration. If enabled, a secure listener enforcing mutual TLS authentication will be started on a different port from the standard plaintext admin server.

This endpoint is secured using identity management, which must be configured separately (because identity management does not depend on remote administration). See the admin/identity config struct.

EXPERIMENTAL: Subject to change.

type ReplacementFunc added in v1.0.4 

type ReplacementFunc func(variable string, val any) (any, error)

ReplacementFunc is a function that is called when a replacement is being performed. It receives the variable (i.e. placeholder name) and the value that will be the replacement, and returns the value that will actually be the replacement, or an error. Note that errors are sometimes ignored by replacers.

type Replacer added in v1.0.4 

type Replacer struct {
	// contains filtered or unexported fields
}

Replacer can replace values in strings. A default/empty Replacer is not valid; use NewReplacer to make one.

func NewEmptyReplacer added in v1.0.4 

func NewEmptyReplacer() *Replacer

NewEmptyReplacer returns a new Replacer, without the global default replacements.

func NewReplacer added in v1.0.4 

func NewReplacer() *Replacer

NewReplacer returns a new Replacer.

func (*Replacer) Delete added in v1.0.4 

func (r *Replacer) Delete(variable string)

Delete removes a variable with a static value that was created using Set.

func (*Replacer) Get added in v1.0.4 

func (r *Replacer) Get(variable string) (any, bool)

Get gets a value from the replacer. It returns the value and whether the variable was known.

func (*Replacer) GetString added in v1.0.4 

func (r *Replacer) GetString(variable string) (string, bool)

GetString is the same as Get, but coerces the value to a string representation as efficiently as possible.

func (*Replacer) Map added in v1.0.4 

func (r *Replacer) Map(mapFunc ReplacerFunc)

Map adds mapFunc to the list of value providers. mapFunc will be executed only at replace-time.

func (*Replacer) ReplaceAll added in v1.0.4 

func (r *Replacer) ReplaceAll(input, empty string) string

ReplaceAll efficiently replaces placeholders in input with their values. All placeholders are replaced in the output whether they are recognized or not. Values that are empty string will be substituted with empty.

func (*Replacer) ReplaceFunc added in v1.0.4 

func (r *Replacer) ReplaceFunc(input string, f ReplacementFunc) (string, error)

ReplaceFunc is the same as ReplaceAll, but calls f for every replacement to be made, in case f wants to change or inspect the replacement.

func (*Replacer) ReplaceKnown added in v1.0.4 

func (r *Replacer) ReplaceKnown(input, empty string) string

ReplaceKnown is like ReplaceAll but only replaces placeholders that are known (recognized). Unrecognized placeholders will remain in the output.

func (*Replacer) ReplaceOrErr added in v1.0.4 

func (r *Replacer) ReplaceOrErr(input string, errOnEmpty, errOnUnknown bool) (string, error)

ReplaceOrErr is like ReplaceAll, but any placeholders that are empty or not recognized will cause an error to be returned.

func (*Replacer) Set added in v1.0.4 

func (r *Replacer) Set(variable string, value any)

Set sets a custom variable to a static value.

func (*Replacer) WithoutFile added in v1.0.4 

func (r *Replacer) WithoutFile() *Replacer

WithoutFile returns a copy of the current Replacer without support for the {file.*} placeholder, which may be unsafe in some contexts.

EXPERIMENTAL: Subject to change or removal.

type ReplacerFunc added in v1.0.4 

type ReplacerFunc func(key string) (any, bool)

ReplacerFunc is a function that returns a replacement for the given key along with true if the function is able to service that key (even if the value is blank). If the function does not recognize the key, false should be returned.

type SinkLog added in v1.0.4 

type SinkLog struct {
	BaseLog
}

SinkLog configures the default Go standard library global logger in the log package. This is necessary because module dependencies which are not built specifically for Kengine will use the standard logger. This is also known as the "sink" logger.

type StderrWriter added in v1.0.4 

type StderrWriter struct{}

StderrWriter writes logs to standard error.

func (StderrWriter) KengineModule added in v1.0.4 

func (StderrWriter) KengineModule() ModuleInfo

KengineModule returns the Kengine module information.

func (StderrWriter) OpenWriter added in v1.0.4 

func (StderrWriter) OpenWriter() (io.WriteCloser, error)

OpenWriter returns os.Stderr that can't be closed.

func (StderrWriter) String added in v1.0.4 

func (StderrWriter) String() string

func (StderrWriter) WriterKey added in v1.0.4 

func (StderrWriter) WriterKey() string

WriterKey returns a unique key representing stderr.

type StdoutWriter added in v1.0.4 

type StdoutWriter struct{}

StdoutWriter writes logs to standard out.

func (StdoutWriter) KengineModule added in v1.0.4 

func (StdoutWriter) KengineModule() ModuleInfo

KengineModule returns the Kengine module information.

func (StdoutWriter) OpenWriter added in v1.0.4 

func (StdoutWriter) OpenWriter() (io.WriteCloser, error)

OpenWriter returns os.Stdout that can't be closed.

func (StdoutWriter) String added in v1.0.4 

func (StdoutWriter) String() string

func (StdoutWriter) WriterKey added in v1.0.4 

func (StdoutWriter) WriterKey() string

WriterKey returns a unique key representing stdout.

type StorageConverter added in v1.0.4 

type StorageConverter interface {
	CertMagicStorage() (certmagic.Storage, error)
}

StorageConverter is a type that can convert itself to a valid, usable certmagic.Storage value. (The value might be short-lived.) This interface allows us to adapt any CertMagic storage implementation into a consistent API for Kengine configuration.

type UsagePool added in v1.0.4 

type UsagePool struct {
	sync.RWMutex
	// contains filtered or unexported fields
}

UsagePool is a thread-safe map that pools values based on usage (reference counting). Values are only inserted if they do not already exist. There are two ways to add values to the pool:

  1. LoadOrStore will increment usage and store the value immediately if it does not already exist.
  2. LoadOrNew will atomically check for existence and construct the value immediately if it does not already exist, or increment the usage otherwise, then store that value in the pool. When the constructed value is finally deleted from the pool (when its usage reaches 0), it will be cleaned up by calling Destruct().

The use of LoadOrNew allows values to be created and reused and finally cleaned up only once, even though they may have many references throughout their lifespan. This is helpful, for example, when sharing thread-safe io.Writers that you only want to open and close once.

There is no way to overwrite existing keys in the pool without first deleting it as many times as it was stored. Deleting too many times will panic.

The implementation does not use a sync.Pool because UsagePool needs additional atomicity to run the constructor functions when creating a new value when LoadOrNew is used. (We could probably use sync.Pool but we'd still have to layer our own additional locks on top.)

An empty UsagePool is NOT safe to use; always call NewUsagePool() to make a new one.

func NewUsagePool added in v1.0.4 

func NewUsagePool() *UsagePool

NewUsagePool returns a new usage pool that is ready to use.

func (*UsagePool) Delete added in v1.0.4 

func (up *UsagePool) Delete(key any) (deleted bool, err error)

Delete decrements the usage count for key and removes the value from the underlying map if the usage is 0. It returns true if the usage count reached 0 and the value was deleted. It panics if the usage count drops below 0; always call Delete precisely as many times as LoadOrStore.

func (*UsagePool) LoadOrNew added in v1.0.4 

func (up *UsagePool) LoadOrNew(key any, construct Constructor) (value any, loaded bool, err error)

LoadOrNew loads the value associated with key from the pool if it already exists. If the key doesn't exist, it will call construct to create a new value and then stores that in the pool. An error is only returned if the constructor returns an error. The loaded or constructed value is returned. The loaded return value is true if the value already existed and was loaded, or false if it was newly constructed.

func (*UsagePool) LoadOrStore added in v1.0.4 

func (up *UsagePool) LoadOrStore(key, val any) (value any, loaded bool)

LoadOrStore loads the value associated with key from the pool if it already exists, or stores it if it does not exist. It returns the value that was either loaded or stored, and true if the value already existed and was loaded, false if the value didn't exist and was stored.

func (*UsagePool) Range added in v1.0.4 

func (up *UsagePool) Range(f func(key, value any) bool)

Range iterates the pool similarly to how sync.Map.Range() does: it calls f for every key in the pool, and if f returns false, iteration is stopped. Ranging does not affect usage counts.

This method is somewhat naive and acquires a read lock on the entire pool during iteration, so do your best to make f() really fast, m'kay?

func (*UsagePool) References added in v1.0.4 

func (up *UsagePool) References(key any) (int, bool)

References returns the number of references (count of usages) to a key in the pool, and true if the key exists, or false otherwise.

type Validator added in v1.0.4 

type Validator interface {
	Validate() error
}

Validator is implemented by modules which can verify that their configurations are valid. This method will be called after Provision() (if implemented). Validation should always be fast (imperceptible running time) and an error must be returned if the module's configuration is invalid.

type WriterOpener added in v1.0.4 

type WriterOpener interface {
	fmt.Stringer

	// WriterKey is a string that uniquely identifies this
	// writer configuration. It is not shown to humans.
	WriterKey() string

	// OpenWriter opens a log for writing. The writer
	// should be safe for concurrent use but need not
	// be synchronous.
	OpenWriter() (io.WriteCloser, error)
}

WriterOpener is a module that can open a log writer. It can return a human-readable string representation of itself so that operators can understand where the logs are going.

Source Files

View all Source files

Directories

Path Synopsis
cmd
kengine
Package main is the entry point of the Kengine application.
 Package main is the entry point of the Kengine application.
filesystems
metrics
testmocks
httpkenginefile
kenginefile
modules
filestorage
kengineevents
kengineevents/eventsconfig
Package eventsconfig is for configuring kengineevents.App with the Kenginefile.
 Package eventsconfig is for configuring kengineevents.App with the Kenginefile.
kenginefs
kenginehttp
kenginehttp/encode
Package encode implements an encoder middleware for Kengine.
 Package encode implements an encoder middleware for Kengine.
kenginehttp/encode/brotli
kenginehttp/encode/gzip
kenginehttp/encode/zstd
kenginehttp/fileserver
kenginehttp/headers
kenginehttp/intercept
kenginehttp/kengineauth
kenginehttp/logging
kenginehttp/map
kenginehttp/proxyprotocol
kenginehttp/push
kenginehttp/requestbody
kenginehttp/reverseproxy
kenginehttp/reverseproxy/fastcgi
kenginehttp/reverseproxy/forwardauth
kenginehttp/rewrite
kenginehttp/standard
kenginehttp/templates
kenginehttp/tracing
kenginepki
kenginepki/acmeserver
kenginetls
kenginetls/distributedstek
Package distributedstek provides TLS session ticket ephemeral keys (STEKs) in a distributed fashion by utilizing configured storage for locking and key sharing.
 Package distributedstek provides TLS session ticket ephemeral keys (STEKs) in a distributed fashion by utilizing configured storage for locking and key sharing.
kenginetls/standardstek
logging
metrics
standard
Package notify provides facilities for notifying process managers of state changes, mainly for when running as a system service.
 Package notify provides facilities for notifying process managers of state changes, mainly for when running as a system service.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.